Spaces:
Running
Running
Create test_streamlit_app.py
Browse files- tests/test_streamlit_app.py +66 -0
tests/test_streamlit_app.py
ADDED
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import unittest
|
2 |
+
from unittest.mock import patch, MagicMock
|
3 |
+
import torch
|
4 |
+
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
5 |
+
import streamlit as st
|
6 |
+
import io
|
7 |
+
|
8 |
+
class TestStreamlitApp(unittest.TestCase):
|
9 |
+
|
10 |
+
@patch("transformers.AutoTokenizer.from_pretrained")
|
11 |
+
@patch("transformers.AutoModelForSequenceClassification.from_pretrained")
|
12 |
+
def test_load_model_success(self, mock_model, mock_tokenizer):
|
13 |
+
# Mock the tokenizer and model loading
|
14 |
+
mock_tokenizer.return_value = MagicMock(spec=AutoTokenizer)
|
15 |
+
mock_model.return_value = MagicMock(spec=AutoModelForSequenceClassification)
|
16 |
+
|
17 |
+
tokenizer, model = load_model("Canstralian/CyberAttackDetection")
|
18 |
+
|
19 |
+
# Assert that the tokenizer and model are not None
|
20 |
+
self.assertIsNotNone(tokenizer)
|
21 |
+
self.assertIsNotNone(model)
|
22 |
+
mock_tokenizer.assert_called_once_with("Canstralian/CyberAttackDetection")
|
23 |
+
mock_model.assert_called_once_with("Canstralian/CyberAttackDetection")
|
24 |
+
|
25 |
+
@patch("transformers.AutoTokenizer.from_pretrained")
|
26 |
+
@patch("transformers.AutoModelForSequenceClassification.from_pretrained")
|
27 |
+
def test_predict_classification(self, mock_model, mock_tokenizer):
|
28 |
+
# Mock the tokenizer and model for inference
|
29 |
+
mock_tokenizer.return_value = MagicMock(spec=AutoTokenizer)
|
30 |
+
mock_model.return_value = MagicMock(spec=AutoModelForSequenceClassification)
|
31 |
+
|
32 |
+
# Simulate model outputs
|
33 |
+
mock_model.return_value.__call__.return_value = MagicMock(logits=torch.tensor([[1.0, 2.0, 3.0]]))
|
34 |
+
|
35 |
+
# Call the prediction function
|
36 |
+
inputs = mock_tokenizer("Test input", return_tensors="pt", padding=True, truncation=True)
|
37 |
+
with torch.no_grad():
|
38 |
+
outputs = mock_model.return_value(**inputs)
|
39 |
+
logits = outputs.logits
|
40 |
+
predicted_class = torch.argmax(logits, dim=-1).item()
|
41 |
+
|
42 |
+
# Assert that the predicted class is correct
|
43 |
+
self.assertEqual(predicted_class, 2) # The class with the highest score (index 2)
|
44 |
+
|
45 |
+
@patch("transformers.AutoTokenizer.from_pretrained")
|
46 |
+
@patch("transformers.AutoModelForSeq2SeqLM.from_pretrained")
|
47 |
+
def test_generate_shell_command(self, mock_model, mock_tokenizer):
|
48 |
+
# Mock the tokenizer and model for shell command generation
|
49 |
+
mock_tokenizer.return_value = MagicMock(spec=AutoTokenizer)
|
50 |
+
mock_model.return_value = MagicMock(spec=AutoModelForSeq2SeqLM)
|
51 |
+
|
52 |
+
# Simulate model output (fake shell command)
|
53 |
+
mock_model.return_value.generate.return_value = torch.tensor([[1, 2, 3, 4]])
|
54 |
+
|
55 |
+
# Simulate text input
|
56 |
+
user_input = "Create a directory"
|
57 |
+
inputs = mock_tokenizer(user_input, return_tensors="pt", padding=True, truncation=True)
|
58 |
+
with torch.no_grad():
|
59 |
+
outputs = mock_model.return_value.generate(**inputs)
|
60 |
+
generated_command = mock_tokenizer.decode(outputs[0], skip_special_tokens=True)
|
61 |
+
|
62 |
+
# Assert the generated command is as expected
|
63 |
+
self.assertEqual(generated_command, "mkdir directory") # Assuming the model generates this
|
64 |
+
|
65 |
+
if __name__ == "__main__":
|
66 |
+
unittest.main()
|